Economic impact modeling
and the New EU Cohesion policy:
The case of the GMR-approach
Attila Varga
University of Pécs
Introduction
• The modern development policy paradigm
– Disappointment in traditional convergence-oriented
policies
– Focus: overall economic growth instead of regional
convergence
– Awareness of the role of geography in growth stimulation
• Two recent approaches:
– Space neutral: to strengthen agglomeration forces in the
economic core with spatially blind interventions – World
Bank (2009)
– Place-based: to utilize growth opportunities in all regions
with integrated regional-specific development
interventions – Barca (2009), OECD (2009)
1
Introduction
• Space-neutral or place-based?
“The debate, then, is not whether policies should be
spatially-blind or place-based, rather how to ensure
these policies enhance each other, are well
synchronised, and do not work against each other.”
Oliveira et al (2014)
2
Introduction
• Important aspects of policy assessment:
– Overall policy impact of a certain policy budget (at
the national and supranational level) depends not
only on the specific instruments applied but also
on the concrete geographic patterns in which
these instruments are deployed regionally
3
Introduction
The role of geography in policy effectiveness
1. Interventions happen at a certain point in space and their impacts are
related to regional structural features (industrial specialization, factor
endowment, agglomeration effects, institutional development).
2. Regional impacts influence other locations (through proximity and
interregional network effects, interregional trade).
3. Interventions lead to a cumulative long run process resulting from labor
and capital migration which further amplify/reduce the initial impacts in the
region and related other locations.
4. Different spatial patterns of interventions might result in significantly
different growth and convergence/divergence patterns.
4
Introduction
A pronounced need towards the incorporation of
the role of geography in policy assessment:
“As policy-makers have limited knowledge on how
and where to intervene, the spatial effects of all
national and supranational policies should be made
explicit. Because there are strong
interdependencies between places, the spatial
consequences of exogenous intervention need to
be assessed and compared.” (Barca 2009, p. 24)
5
Introduction
• The related role of economic models:
– Specially constructed economic models could provide
information for policymakers in the assessment of the
spatial consequences of various policy shocks.
– These models could also provide information in
comparing the effects of different projects following
particular geographic and instrumental combinations
of a set of instruments.
6
Introduction
• Economic impact models and modern
development policy:
– Traditional macroeconomic models (econometric,
CGE, DSGE)
• Do not reflect geographical effects (agglomeration,
migration, interregional trade, spillovers, regional
economic structures)
– Regional and interregional CGE models
• Do not reflect the influence of macroeconomic factors
(macroeconomic conditions/policies) on the
effectiveness of regional interventions
7
Introduction
• Search for new modeling approaches (MASST,
GMR-type models (GMR-Hungary, GMREurope, RHOMOLO), system dynamic
approach)
• This presentation:
– classifies the challenges towards economic
modeling;
– illustrates a reflection to the challenges by the
GMR- Europe model.
8
Outline
• Modeling challenges
• The GMR-approach
• Reflections to the challenges in the GMREurope model
• Policy simulation examples
9
Modeling challenges
• Step 1: Modeling policy impact on regional
technological progress
– Incorporation of the mechanisms discovered in
the geography of innovation literature: local /
global knowledge flows, different agglomeration
effects (MAR or Jacobs, related variety),
entrepreneurship
– Modeling possibilities:
• knowledge production function (Varga et al 2013)
• evolutionary techniques (Faggiolo, Dosi 2003)
10
Modeling challenges
• Step 2: Modeling the transmission of the
technology impact to economic variables
– Productivity and variety impacts (Saviotti, Pyka
2003)
– What growth theories offer:
• Romer 1990 – productivity impact at the end
• Aghion, Howitt 1998: limited variety impact
• Evolutionary theories get closer to formulating variety
effects (Saviotti, Pyka 2003, Faggiolo, Dosi 2003)
– Technical difficulties, problems with regional data
11
Modeling challenges
• Step 3: Modeling spatiotemporal dynamics of
economic growth
– Spatiotemporal dynamics modeling: accounting for both
the direct indigenous extension of production factors and
their changing spatial patterns via factor migration
– Option 1: Spatiotemporal dynamics modeled at the level of
regions
• Forward looking expectations (Bröcker, Korzhenevych 2011)
• Alternative investment and saving behavior (Ivanova et al 2007)
– Option 2: Spatiotemporal dynamics modeled separately in
macro and regional models (Varga et al. 2011)
12
Modeling challenges
• Step 4: Macro impact integration
– Impacts of macroeconomic framework conditions
– New and open area of research (Varga et al. 2011)
13
The GMR-approach
• provides an integrated framework for
– a comparative assessment of alternative regional
policies on the basis of their overall impact on
national growth;
– the incorporation of macroeconomic factors in the
impact assessment of regional interventions.
14
GMR: Geographic Macro and Regional modeling
15
Why “geographic”?
• To emphasize that the role of geography in
development policy impacts is placed at the
core of the models.
16
Why „regional”?
• Spatial reference unit where interventions
happen
17
Why „macro”?
• Macro level impacts form the basis of the
overall evaluation of alternative regional
interventions.
• Awareness of the role of macroeconomic
factors in regional policy effectiveness.
18
The GMR approach:
Antecedens and applications
• Antecedents:
– Links to theory: Acs-Varga 2002
– Empirical modeling framework (Varga 2006)
– The EcoRet model (Schalk, Varga 2004, Varga, Schalk
2004)
– The GMR-Hungary model (Varga, Schalk, Koike, Járosi,
Tavasszy 2008; Járosi, Koike, Thissen, Varga 2010)
– Dynamic KPF model for EU regions (Varga, Pontikakis,
Chorafakis, 2013)
– GMR-EU (Varga, Járosi, Sebestyén 2011; Varga,Törma
2011)
– GMR-HUNGARY II (Varga, Járosi, Sebestyén, 2013)
• Applications: Cohesion Policy impact studies for the
European Commission (DG Regio) and the Hungarian
government; FP6 impact study
Structure of GMR models
– a regional Total Factor Productivity (TFP)
block
– a regional Spatial Computable General
Equilibrium (SCGE) block
– a macroeconomic (MACRO) block
20
Reflections to challenges in the GMREurope model
• Step 1: Modeling policy impact on technological
progress (TFP block)
– Spatialised extension of the Romer 1990 knowledge
production model incorporating several elements of
the findings in the geography of innovation literature
(Varga et al 2013, Sebestyén, Varga 2013):
• Dynamic agglomeration effects
• Interregional knowledge flows (co-patenting, co-publication
network effects)
• Interregional spillovers – with no specific mechanisms
identified (spatial econometrics)
21
R&D productivity publications
Regional
attractiveness: R&D
R&D
Regional
attractiveness:
knowledge industries
Knowledge industry concentration
R&D productivitypatenting
Interregional
research networks
Patenting in
proximate regions
Patenting
National technological
development
Regional technological
development
Social
capital
Human capital
Industrial
concentration
TFP
Technological development in
proximate regions
Figure 1: The estimated regional dynamics of innovation policies in the TFP block of
the GMR-Europe model
22
Reflections to challenges in the GMREurope model
• Step 2. Modeling the transmission of the
technology impact to economic variables (TFP
block)
– Technological ideas channeled through their TFP
effects:
aTFP1SOCKAPi,t-k
TFPi,t = aTFP0 HCAPi,t-k
aTFP 2 ln( Li,t-k AREAi )
Ai,t-k
aTFP 3
W _ Ai,t-k
23
Reflections to challenges in the GMREurope model
• Steps 3 and 4: Modeling spatiotemporal
dynamics of economic growth and macro impact
integration (SCGE and MACRO blocks)
– Step 3a: Short run regional equilibrium (given K and L,
no migration) – system of regional CGE models
– Step 3b: Spatial dynamics via migration and altered
regional TFP – in the system of regional CGE models
– Step 3c: Dynamic aggregate impacts on K and L: the
macro model
– Step 4: Aggregated impacts distributed over regions
24
!!!
Policy
Spatiotemporal dynamics
Impacts
MACRO block
Changes in aggregate
K and L
DTFPN,t
Regional SCGE block
Spatial equilibrium with
given KN and LN
DLi,t
!
R&D, human
capital, physical
accessibility
Macroeconomic
(TFP, K, L, Y, inflation,
wages, etc.)
DKN,t DLN,t
Regional
(TFP, K, L, wages, prices)
DTFPi,t
Regional TFP block
Policy-induced changes
in TFP
!
Figure 2: Regional and macro impacts of regionally implemented innovation policies
in the GMR-Europe model
25
26
0,3000%
3,5000%
0,2500%
3,0000%
2,5000%
0,2000%
2,0000%
0,1500%
1,5000%
0,1000%
1,0000%
0,0500%
0,5000%
0,0000%
0,0000%
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Orig
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Agglomeration
The Agglomeration effect: Greece
0,9000%
Orig
Agglomeration
The Agglomeration effect: Portugal
1,4000%
0,8000%
1,2000%
0,7000%
1,0000%
0,6000%
0,5000%
0,8000%
0,4000%
0,6000%
0,3000%
0,4000%
0,2000%
0,2000%
0,1000%
0,0000%
0,0000%
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Orig
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Agglomeration
The Agglomeration effect: Czech Republic
Orig
The Agglomeration effect: Hungary
1,4000%
0,1200%
1,2000%
0,1000%
1,0000%
Agglomeration
0,0800%
0,8000%
0,0600%
0,6000%
0,0400%
0,4000%
0,0200%
0,2000%
0,0000%
0,0000%
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Orig
Agglomeration
The Agglomeration effect: Slovak Republic
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Orig
Agglomeration
The Agglomeration effect: Euro zone + CZ, HU, SK
Figure 6: Results of the Agglomeration and concentration scenario
Percentage differences between scenario and baseline GDP values.
27
3,5000%
0,3000%
3,0000%
0,2500%
2,5000%
0,2000%
2,0000%
0,1500%
1,5000%
0,1000%
1,0000%
0,0500%
0,5000%
0,0000%
0,0000%
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Orig
Donut
Orig
Donut
The Donut effect: Greece
The Donut effect: Portugal
0,9000%
1,4000%
0,8000%
1,2000%
0,7000%
1,0000%
0,6000%
0,5000%
0,8000%
0,4000%
0,6000%
0,3000%
0,4000%
0,2000%
0,2000%
0,1000%
0,0000%
0,0000%
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Orig
Donut
The Donut effect: Czech Republic
1,6000%
Orig
Donut
The Donut effect: Hungary
0,1200%
1,4000%
0,1000%
1,2000%
0,0800%
1,0000%
0,8000%
0,0600%
0,6000%
0,0400%
0,4000%
0,0200%
0,2000%
0,0000%
0,0000%
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
Orig
Donut
The Donut effect: Slovak Republic
Orig
Donut
The Donut effect: Euro zone + CZ, HU, SK
Figure 5: Results of the Donut scenario
Percentage differences between scenario and baseline GDP values.
28